This is to provide an update of the current status of the phase 1 of the
smart grid project (note: description of this project can be found in the
thread "Smart Grid Project : ML Predictions").

In the phase 1 we use  WSO2 ML to predict the next market clearing price
for the next 15 min period. The clearing price prediction is done by first
training the machine learning algorithm using the weather data. The
training data set contains clearing prices, wind speed and solar radiation
that was observed during the previous market clearing. Using this model and
the weather data of the next 15 min, we predict the clearing price for the
15 min period. This information is then sent to the  GridLab-D (power
distribution system simulator). GridLab-D is configured to simulate 629
houses, large and small wind turbines and a set of solar panels connected
together and to power grid via IEEE-12 distribution feeder. The consumers
(houses etc) and generators (i.e. wind turbines etc) within the GridLab-D
use the predicted clearing price to decide how they bid. The overall
objective is to minimize the costs associated with drawing additional
energy (i.e. from main grid), due to temporary variations in the energy
generated by renewable resources.

1) To predict the next clearing price we have tried different ML algorithms

@Sanjaya could you please provide the summary of results (accuracy etc) we
got under different machine learning algorithms (i.e. random forest
regression etc)

2) In GridLab-D we compute how much energy is saved and the cost savings
@Nihla  could you please provide the cost/energy savings we got as a result
of using ML


Thanks

-- 
Malith Jayasinghe



WSO2, Inc. (http://wso2.com)
Email   : [email protected]
Mobile : 0770704040
Lean . Enterprise . Middleware
_______________________________________________
Architecture mailing list
[email protected]
https://mail.wso2.org/cgi-bin/mailman/listinfo/architecture

Reply via email to